POWER BREAKTHROUGHS IN HEALTHCARE AND LIFE SCIENCES

• Kimberly Powell, Vice President of Healthcare, NVIDIA

Hospitals, operating rooms, genomic sequencing centers, and pharmaceutical companies generate massive amounts of data, creating the opportunity to build AI models, platforms, and robotic systems that can predict, comprehend, learn, and act. Get the latest on how life science researchers, developers, and medical device makers are using NVIDIA Clara™ to invent breakthroughs in healthcare delivery and drug discovery.

• Ola Engkvist, Head of Molecular AI, AstraZeneca

Artificial intelligence has become impactful during the last few years in chemistry and the life sciences, pushing scientific boundaries forward as exemplified by the recent success of AlphaFold2. In this presentation, I’ll provide an overview of how AI has impacted drug design in the last few years, where we are now, and what progress we can reasonably expect in the coming years. The presentation will have a focus on deep learning-based molecular de novo design; however, aspects of synthesis prediction, molecular property predictions, and chemistry automation will also be covered.

• Caroline Chung, Chief Data Officer, MD Anderson Cancer Center

With the exponential growth in volume of medical imaging data, the mounting need for AI to help clinical teams handle and optimally use all the available imaging is apparent. However, developing AI tools that have successfully been translated into the clinical setting has been relatively slow and cautious. Challenges faced in development and clinical adoption may have solutions that lie within the imaging-associated metadata, as the impact of factors such as image quality and generalizability of AI models may be reflected in the rich contextual information that surrounds an imaging exam.

• Euan Ashley, Professor of Medicine and Genetics, Stanford University

This session will look at how a collaborative team recently set the Guinness World Record for the fastest DNA sequencing technique while making medical diagnoses in under eight hours from whole genome sequencing in critical care patients at Stanford hospitals. We’ll explore how something that used to take years and cost millions of dollars can now be realized in hours in modern hospital settings and outline the computational and algorithmic infrastructure that supported the advance.

• Mohammed AlQuraishi, Professor, Columbia University

Deep learning methods that predict structures with accuracy rivaling experimental methods have revolutionized protein structure prediction. Such methods require substantial amounts of compute to train, however, and code and data infrastructure to learn new models tailored to new tasks.

In this session, learn about OpenFold, a new open-source platform for training leading protein structure prediction models. OpenFold provides the code and data necessary to train new models from scratch, as well as pr-trained models under a permissive license for broad use by academia and industry.